fixes

Overview

Current version: 0.7.1

Note
By default, the fixes package assumes time is a regularly spaced numeric variable (e.g., year = 1995, 1996, …).
If your time variable is irregular or non-numeric (e.g., Date type), set time_transform = TRUE to automatically convert it to a sequential index within each unit.
For unit-specific treatment timing, set staggered = TRUE.

The fixes package is designed for convenient event study analysis and plotting, particularly useful for visualizing parallel trends and dynamic effects in two-way fixed effects (TWFE) difference-in-differences (DID) research.

Key Functions:

run_es() — Takes a data frame, generates lead/lag dummies, and fits the event study regression. Supports fixed effects, covariates, clustering, staggered timing, weights, custom baseline, and multiple confidence intervals.
plot_es() — Plots event study results using ggplot2 with flexible options: ribbon or error bars, choice of CI level, and theme customization.
plot_es_interactive() — Creates interactive event study plots using plotly with hover tooltips displaying coefficients, confidence intervals, standard errors, and p-values.

Installation

Install from CRAN:

install.packages("fixes")

Or with pak:

pak::pak("fixes")

For the latest development version from GitHub:

pak::pak("yo5uke/fixes")

How to use

First, load the library.

library(fixes)

Data frame requirements

run_es() expects a panel data frame with at least:

Unit identifier (e.g., individual, firm, region)
Treatment indicator (0/1 or TRUE/FALSE)
Time variable (numeric or Date)
Outcome variable (continuous)

For staggered adoption (staggered = TRUE), include a variable specifying unit-specific treatment timing (e.g., “treatment_year”).

Example data

Widely used panel datasets include:

did::sim_dt(): simulated panel for DiD tutorials
fixest::base_stagg: a built-in dataset for staggered adoption

df1 <- fixest::base_did      # Basic DiD
df2 <- fixest::base_stagg    # Staggered treatment

y	x1	id	period	post	treat
2.8753063	0.5365377	1	1	0	1
1.8606527	-3.0431894	1	2	0	1
0.0941652	5.5768439	1	3	0	1
3.7814749	-2.8300587	1	4	0	1
-2.5581996	-5.0443544	1	5	0	1
1.7287324	-0.6363849	1	6	1	1

	id	year	year_treated	time_to_treatment	treated	x1	y
2	90	1	2	-1	1	-1.0947021	0.0172297
3	89	1	3	-2	1	-3.7100676	-4.5808453
4	88	1	4	-3	1	2.5274402	2.7381717
5	87	1	5	-4	1	-0.7204263	-0.6510307
6	86	1	6	-5	1	-3.6711678	-5.3338166
7	85	1	7	-6	1	-0.3152137	0.4956263

`run_es()`

The main event study function. All key arguments below:

Argument	Description
`data`	Data frame to be used.
`outcome`	Outcome variable. Can be specified as a raw variable or a transformation (e.g., `log(y)`). Provide it unquoted.
`treatment`	Dummy variable indicating the treated units. Provide it unquoted. Accepts both `0/1` and `TRUE/FALSE`.
`time`	Time variable. Provide it unquoted.
`timing`	The time at which the treatment occurs. If `staggered = FALSE`, this should be a scalar (e.g., `2005`). If `staggered = TRUE`, provide a variable (column) indicating the treatment time for each unit.
`fe`	Fixed effects to control for unobserved heterogeneity. Must be a one-sided formula (e.g., `~ id + year`).
`lead_range`	Number of pre-treatment periods to include (e.g., 3 = `lead3`, `lead2`, `lead1`). Default is `NULL`, which automatically uses the maximum available lead range.
`lag_range`	Number of post-treatment periods to include (e.g., 2 = `lag0` (the treatment period), `lag1`, `lag2`). Default is `NULL`, which automatically uses the maximum available lag range.
`covariates`	Additional covariates to include in the regression. Must be a one-sided formula (e.g., `~ x1 + x2`).
`cluster`	Specifies clustering for standard errors. Can be a character vector (e.g., `c("id", "year")`) or a formula (e.g., `~ id + year`, `~ id^year`).
`weights`	Optional weights to be used in the regression. Provide as a one-sided formula (e.g., `~ weight`).
`baseline`	Relative time value to be used as the reference category (default: `-1`). For both `classic` and `sunab` methods, this period is included in results with zero estimates. Must be within the specified lead/lag range.
`interval`	Time interval between observations (e.g., `1` for yearly data, `5` for 5-year intervals).
`time_transform`	Logical. If `TRUE`, converts the `time` variable into a sequential index (1, 2, 3, …) within each unit. Useful for irregular time (e.g., Date). Default is `FALSE`.
`unit`	Required if `time_transform = TRUE`. Specifies the panel unit identifier (e.g., `firm_id`).
`staggered`	Logical. If `TRUE`, allows for unit-specific treatment timing (staggered adoption). Default is `FALSE`.
`method`	Estimation method: `"classic"` (default) uses factor expansion with `fixest::i()`, or `"sunab"` for staggered-robust estimation using Sun & Abraham (2021) decomposition.
`conf.level`	Numeric vector of confidence levels (e.g., `c(0.90, 0.95, 0.99)`; default: `0.95`).
`vcov`	Variance-covariance type (default: `"HC1"`). Accepts any `fixest::vcov()` type (e.g., `"HC3"`, `"CR2"`, `"iid"`).

Example: basic event study

event_study <- run_es(
  data       = df1,
  outcome    = y,
  treatment  = treat,
  time       = period,
  timing     = 5,  # Treatment occurs at period 5
  fe         = ~ id + period,
  lead_range = 3,
  lag_range  = 3,
  cluster    = ~ id,
  baseline   = -1,
  interval   = 1,
  conf.level = c(0.90, 0.95, 0.99)
)

# View summary
print(event_study)

## Event Study Result (fixes)
##   N: 1080  | Units: NA  | Treated units: 1080  | Never-treated: NA 
##   FE: id + period
##   VCOV: HC1  | Cluster: id 
##   Method: classic  | lead_range: 3  lag_range: 3  baseline: -1

fe must be a one-sided formula (e.g., ~ firm_id + year).
cluster can be a one-sided formula or a character vector.

Plotting Results

Example usage

# Basic plot with ribbon (default: 95% CI)
plot_es(event_study)

# Plot with error bars
plot_es(event_study, type = "errorbar", ci_level = 0.95)

# Customized plot
plot_es(event_study, type = "ribbon", ci_level = 0.99, theme_style = "minimal") +
  ggplot2::ggtitle("Event Study: 99% Confidence Interval")

Staggered Treatment with `sunab`

For staggered adoption designs with unit-varying treatment timing, use method = "sunab" for the Sun & Abraham (2021) estimator:

# Using fixest::base_stagg example data
event_study_sunab <- run_es(
  data       = df2,
  outcome    = y,
  treatment  = treated,
  time       = year,
  timing     = year_treated,
  fe         = ~ id + year,
  staggered  = TRUE,
  method     = "sunab",
  lead_range = 3,
  lag_range  = 3,
  cluster    = ~ id
)

# View summary
print(event_study_sunab)

## Event Study Result (fixes)
##   N: 950  | Units: NA  | Treated units: 950  | Never-treated: NA 
##   FE: id + year
##   VCOV: HC1  | Cluster: id 
##   Method: SUNAB (staggered-safe)

# Plot sunab results
plot_es(event_study_sunab)

Note (v0.7.1+): The baseline parameter now works for both classic and sunab methods, and results are properly filtered to lead_range and lag_range.

Interactive Plots

Create interactive event study plots with hover tooltips (requires the plotly package):

# Interactive plot with hover information
plot_es_interactive(event_study, ci_level = 0.95)

Hover over points to see relative time, point estimates, confidence intervals, standard errors, and p-values.

Planned Features

Support for staggered = TRUE with time_transform = TRUE
Allow timing to accept original time values (e.g., Date), not just index

Debugging and Contributions

If you find an issue or want to contribute, please use the GitHub Issues page.

Happy analyzing!🥂